This reverts commit r206707, reapplying r206704. The preceding commit
to CalcSpillWeights should have sorted out the failing buildbots.
<rdar://problem/14292693>
llvm-svn: 206766
Our build is still getting this test to fail with the presence of const in
'(Foo &const) y = 0x' and '(Foo &&const) z = 0x'. This change simply changes
the match substrings to ignore the portion after the final '&' and before the
closing ')'. This should cover the presence (or absence) of the const qualifier.
llvm-svn: 206762
The underlying issue was actually a Clang bug, now fixed. The test now
reports XPASS for me locally and on the buildbot.
llvm.org/pr17183 (LLDB)
llvm.org/pr18950 (Clang)
llvm-svn: 206761
Perhaps these should be @expectedFailureDarwin instead of
@unittest2.expectedFailure (applying to all hosts); I'm not aware of
the details in the rdars.
Just add a comment for now, for the benefit of anyone investigating
FreeBSD test issues in the future.
rdar://9980907
rdar://15367233
llvm-svn: 206760
ocasionally it fails with a slightly different report:
WARNING: ThreadSanitizer: signal handler spoils errno (pid=3674)
#0 MyHandler(int, siginfo*, void*)
...
#4 __tsan_free_hook
#5 main signal_errno.cc:40
llvm-svn: 206754
The additional "const" breaks the FreeBSD buildbot, and does not appear
to be due to a LLVM or Clang change. Revert the change while
investigating further.
This reverts revision 206619.
llvm-svn: 206751
We normally don't drop functions from the C API's, but in this case I think we
can:
* The old implementation of getFileOffset was fairly broken
* The introduction of LLVMGetSymbolFileOffset was itself a C api breaking
change as it removed LLVMGetSymbolOffset.
* It is an incredibly specialized use case. The only reason MCJIT needs it is
because of its odd position of being a dynamic linker of .o files.
llvm-svn: 206750
LazyCallGraph analysis framework. Wire it up all the way through the opt
driver and add some very basic testing that we can build pass pipelines
including these components. Still a lot more to do in terms of testing
that all of this works, but the basic pieces are here.
There is a *lot* of boiler plate here. It's something I'm going to
actively look at reducing, but I don't have any immediate ideas that
don't end up making the code terribly complex in order to fold away the
boilerplate. Until I figure out something to minimize the boilerplate,
almost all of this is based on the code for the existing pass managers,
copied and heavily adjusted to suit the needs of the CGSCC pass
management layer.
The actual CG management still has a bunch of FIXMEs in it. Notably, we
don't do *any* updating of the CG as it is potentially invalidated.
I wanted to get this in place to motivate the new analysis, and add
update APIs to the analysis and the pass management layers in concert to
make sure that the *right* APIs are present.
llvm-svn: 206745
became empty. This would manifest later as an assert failure due to
a non-empty list map but an empty result map. This doesn't easily
manifest with just the module pass manager and the function pass
manager, but the next commit will add the CGSCC pass manager that hits
this assert immediately.
llvm-svn: 206744
Generating BZHI in the variable mask case, i.e. (and X, (sub (shl 1, N), 1)),
was already supported, but we were missing the constant-mask case. This patch
fixes that.
<rdar://problem/15480077>
llvm-svn: 206738
file. This will make it easy to scale up the number of passes supported.
Currently, it just supports the function and module transformation
passes that were already supported in the opt tool explicitly.
llvm-svn: 206737
The lldb-gdbserver tests are skipped if the lldb-gdbserver exe cannot be found (currently only built for Linux and FreeBSD). The lldb-gdbserver exe is found by the LLDB_EXEC environment variable, using the same dir as the lldb exe, and then replacing lldb with lldb-gdbserver.
llvm-svn: 206731
Unlike the standard AAPCS64 ABI, variadic arguments are always passed on the
stack with the Darwin ABI, and this was not being considered when deciding
whether to expand HFA/HVA arguments in a call. An HFA argument with a "float"
base type was being expanded into separate "float" arguments, each of which
was then extended to a double, resulting in a serious mismatch from what is
expected by the va_arg implementation. <rdar://problem/15777067>
llvm-svn: 206729
It could even be made non-virtual if it weren't for bad compiler
warnings.
This demonstrates that ArgList objects aren't destroyed polymorphically
and possibly that they aren't even used polymorphically. If that's the
case, it might be possible to refactor the two ArgList types more
separately and simplify the Arg ownership model. *continues
experimenting*
llvm-svn: 206727
Ben Langmuir